Paper builds a 24-case dataset of MCP security vulnerabilities
Original title: Tracing MCP Security Vulnerabilities: Constructing and Analyzing an MCP Security Vulnerability Dataset
Authors say 87.5 percent are Critical or High, and injection plus path traversal account for 66.7 percent.
The paper constructs a curated Model Context Protocol vulnerability dataset from CVE disclosures. From 119 MCP-related records retrieved through February 2026, filtering produced 24 validated vulnerabilities. The authors report recurring issues in input validation, access control, and protocol-level security, and illustrate patterns in six case studies. They release the dataset, reproducibility tooling, and a maintenance workflow.
Unconfirmed: Severity and weakness shares describe the authors' 24-case snapshot, not all MCP software.
- A sampled exclusion audit reached 96.7 percent inter-rater agreement.
- Filtering kept all known high-severity MCP cases from public advisories.
- The methodology explicitly addresses MCP acronym ambiguity.
- The release is framed as a foundation for later measurement and defense research.
From the original article
1 min read at arxiv.orgAs a core protocol for integrating large language models (LLMs) with external data sources and tools, the Model Context Protocol (MCP) has quickly become a key interface in emerging AI agent systems. As the MCP ecosystem expands, reported vulnerabilities have also increased, creating a growing software-security attack surface.
By Xu He · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.